Prevalence of post-traumatic stress disorder and common mental disorders in health-care workers in England during the COVID-19 pandemic: a two-phase cross-sectional study

of UK health-care workers used diagnostic interviewing to estimate prevalence of PTSD (44%) and depression (39%) in a sample of 103 front-line health-care workers in England, only interviewing individuals who scored above clinical cutoffs on the PTSD Checklist for DSM-5

and Patient Health Questionnaire-9 screening tools using the Structured Clinical Interview for DSM-5.

The study did not calculate the true population prevalence and prevalence estimates were markedly higher than those found in our study, which might be explained by the use of different diagnostic tools, differences in sample characteristics, and the use of convenience sampling. The sample used by Wild and colleagues consisted primarily of front-line ambulance staff and nurses, recruited from four NHS Hospitals and Ambulance Trusts, limiting generalisability to the wider population of health-care workers. The authors found that most participants diagnosed with PTSD reported an index event that happened before the COVID-19 pandemic, whereas major depressive disorder symptoms seemed to have developed over the course of the pandemic.

We were unable to explore time of onset for the mental health outcomes of interest or the index event for PTSD in this study.

We also included both clinical and non-clinical staff in recognition of the burden that all health-care workers have faced during the pandemic. Acute and mental health NHS Trusts were included in NHS CHECK cohort, therefore exposures to some of the challenges associated with the pandemic would have been experienced differently, for example, by clinical staff working in an ICU, or emergency department, when compared with staff working on an acute psychiatric inpatient unit. Our research could be used in conjunction with findings from other studies that identify which particular groups of health-care workers might be at an increased risk of mental disorders, similar to previous work published by the NHS CHECK team,

which showed that nurses, younger health-care workers, women, and individuals exposed to morally injurious events were at increased risk. The prevalence estimates reported in these studies are likely to be overestimations due to the use of screening tools; however, the identification of risk groups would remain valid.

Previous research has shown that people with mental disorders are less likely to take part in research, and this might have led to an underestimation of our prevalence estimates.

Furthermore, individuals from minority ethnic backgrounds were less likely to have completed the PCL-6 or GHQ-12 during the NHS CHECK baseline assessment, possibly resulting in a selection bias, despite targeted recruitment efforts to increase the number of participants from a minority ethnic background for the diagnostic interviews. It is important to note that the confidence interval for the population prevalence estimate of PTSD was wide due to a small sample size, resulting in a less precise prevalence estimate. A larger number of participants had missing values on the PCL-6 than the GHQ-12 since the PCL-6 was only included in the second, optional part of the baseline questionnaire instead of the first, compulsory part of the baseline questionnaire. We offered a short and long version of this baseline questionnaire taking into consideration pressures of working in a health-care setting during the pandemic and participant burden.

In such studies, prevalence estimates of mental disorders are higher than occupation-specific prevalences extracted from larger true population studies in which occupation is collected as an incidental variable. This has been supported by more research on the impact of COVID-19, where population studies reported lower prevalence estimates in health-care workers than did surveys of health-care workers specifically. Use of diagnostic interviews might reduce this systematic bias by providing a more rigorous assessment of mental disorder, but this is unlikely to eliminate bias completely. Furthermore, no data were available from assessment points before the COVID-19 pandemic, thus we are unable to draw conclusions as to whether the prevalence of common mental disorders and PTSD in this population has changed since the start of the pandemic; the study was also not sufficiently powered to examine differences between subgroups (such as clinical and non-clinical staff, type of NHS Trust, or clinical speciality) due to careful consideration about what was feasible within the time and resources available. A larger study could support or refute the risk factors highlighted in previous research.

We did not directly assess individuals’ need for treatment and cannot be certain what proportion of individuals might benefit from formal treatment. Another limitation was the time lag between the baseline assessment of NHS CHECK for common mental disorders and PTSD and the diagnostic interview study (mean 265 days [SD 8]). Symptoms for some participants might have naturally eased over time, whereas other participants could have developed symptoms of a mental disorder. The longitudinal data of NHS CHECK indicate that despite slight variations in the screening prevalence of common mental disorders and PTSD, depending on the timing of the follow-up assessments, these prevalences remained relatively stable.

Taken together, we believe that our results indicate a true overestimation of common mental disorders and PTSD prevalence when using screening measures compared with diagnostic interviews.

considering the known association between diagnosable mental disorders and poor workplace functioning, we suggest that it might be helpful to provide treatment promptly for health-care workers with diagnosable mental disorders.

This approach is likely to both benefit health-care workers themselves and ensure quality of care for patients by maintaining a well functioning workforce. More broadly, researchers seeking to assess psychiatric symptoms using self-report screening tools in novel contexts should carefully consider cutoff scores on screening tools, and ideally complete further validation work to correctly calibrate measures to be appropriately sensitive. It can be unhelpful to report results from self-report tools since they might cause alarm and inappropriate allocation of scarce resources. Additionally, the variation in important prognostic indicators (past mental illness, urbanicity of clinical service provision, length of clinical contact) needs to be investigated if timely resources and treatments are to be provided to a population with high prevalence of mental distress. Further longitudinal research should be carried out to ascertain whether estimates of common mental disorders and PTSD among health-care workers exist before they start their role, are sustained during employment, or decrease over time.